In [1]:
import librosa
import os
In [2]:
import mir_eval
In [3]:
from collections import OrderedDict
In [4]:
import pandas as pd
import numpy as np
np.set_printoptions(precision=3)
pd.set_option('precision', 4, "display.max_rows", 999)
In [5]:
def make_beatles_corpus(iso_path):
# Beat files
beat_path = os.path.join(iso_path, 'beat')
annotations = librosa.util.find_files(beat_path, ext='txt')
audio = [ann.replace('/beat/', '/audio/').replace('.txt', '.flac') for ann in annotations]
data = []
for aud, ann in zip(audio, annotations):
if os.path.exists(aud) and os.path.exists(ann):
data.append((aud, ann))
return pd.DataFrame(data=data, columns=['audio', 'annotation'])
In [6]:
def make_smc_data(smc_path):
# Beat files
beat_path = os.path.join(smc_path, 'SMC_MIREX_Annotations')
annotations = librosa.util.find_files(beat_path, ext='txt')
# Audio files
audio_path = os.path.join(smc_path, 'SMC_MIREX_Audio')
audio = librosa.util.find_files(audio_path, ext='wav')
data = zip(audio, annotations)
return pd.DataFrame(data=data, columns=['audio', 'annotation'])
In [7]:
def make_output_path(base, outpath):
root = os.path.splitext(base)[0]
output = os.path.join(outpath, os.path.extsep.join([root, 'json']))
return output
In [8]:
def analyze(dframe, outpath='/home/bmcfee/git/librosa_parameters/data/beat/'):
index = dframe.index[0]
base = os.path.basename(dframe['audio'][index])
outfile = make_output_path(base, outpath)
if os.path.exists(outfile):
print 'Cached {}'.format(base)
data = pd.read_json(outfile, orient='records')
return data
else:
print 'Processing {}'.format(base)
# Load the truth
ref_times = pd.read_table(dframe['annotation'][index], header=None, sep='\s+',
usecols=[0], error_bad_lines=False)[0].values
# Load the audio
sr = 22050
y, _ = librosa.load(dframe['audio'][index], None)
y = librosa.resample(y, _, sr, res_type='sinc_fastest')
# Construct the output container
results = []
# Onset strength parameters
for fmax in [8000, 11025]:
for n_mels in [32, 64, 128]:
S = librosa.feature.melspectrogram(y=y, sr=sr, fmax=fmax, n_mels=n_mels)
S = librosa.logamplitude(S)
for aggregate in [np.mean, np.median]:
# Compute the onset detection function
oenv = librosa.onset.onset_strength(S=S, sr=sr,
aggregate=aggregate)
# Tempo estimator parameters
for ac_size in [2, 4, 8]:
for std_bpm in [0.5, 1.0, 2.0]:
tempo = librosa.beat.estimate_tempo(oenv,
sr=sr,
ac_size=ac_size,
std_bpm=std_bpm)
for tightness in [50, 100, 400]:
# Evaluate the predictions
params = {'aggregate': aggregate.__name__,
'fmax': fmax,
'n_mels': n_mels,
'ac_size': ac_size,
'std_bpm': std_bpm,
'tightness': tightness}
_, beats = librosa.beat.beat_track(sr=sr,
onset_envelope=oenv,
trim=False,
tightness=tightness,
bpm=tempo)
est_times = librosa.frames_to_time(beats, sr=sr)
scores = mir_eval.beat.evaluate(ref_times, est_times)
cont = OrderedDict(index=index)
cont.update(params)
cont.update(scores)
results.append(cont)
# Blow away the cache
#librosa.cache.clear()
data = pd.DataFrame.from_dict(results, orient='columns')
data.to_json(outfile, orient='records')
return data
In [9]:
def analyze_corpus(corpus):
results = None
for idx in corpus.index:
new_results = analyze(corpus.loc[[idx]])
if results is None:
results = new_results
else:
results = pd.concat([results, new_results])
return results
In [10]:
from joblib import Parallel, delayed
In [11]:
def p_analyze_corpus(corpus, n_jobs=3):
results = None
dfunc = delayed(analyze)
results = Parallel(n_jobs=n_jobs, verbose=10)(dfunc(corpus.loc[[idx]])
for idx in corpus.index)
return pd.concat(results)
In [12]:
smc_data = make_smc_data('/home/bmcfee/data/SMC_Mirex/')
In [13]:
smc_results = p_analyze_corpus(smc_data)
[Parallel(n_jobs=3)]: Done 1 out of 35 | elapsed: 0.1s remaining: 2.8s
[Parallel(n_jobs=3)]: Done 21 out of 217 | elapsed: 3.6min remaining: 33.6min
[Parallel(n_jobs=3)]: Done 43 out of 217 | elapsed: 6.5min remaining: 26.4min
[Parallel(n_jobs=3)]: Done 65 out of 217 | elapsed: 9.0min remaining: 21.0min
[Parallel(n_jobs=3)]: Done 87 out of 217 | elapsed: 11.7min remaining: 17.4min
[Parallel(n_jobs=3)]: Done 109 out of 217 | elapsed: 14.6min remaining: 14.5min
[Parallel(n_jobs=3)]: Done 131 out of 217 | elapsed: 17.2min remaining: 11.3min
[Parallel(n_jobs=3)]: Done 153 out of 217 | elapsed: 19.9min remaining: 8.3min
[Parallel(n_jobs=3)]: Done 175 out of 217 | elapsed: 22.6min remaining: 5.4min
[Parallel(n_jobs=3)]: Done 197 out of 217 | elapsed: 25.6min remaining: 2.6min
[Parallel(n_jobs=3)]: Done 217 out of 217 | elapsed: 28.2min finished
Cached SMC_002.wav
Cached SMC_001.wav
Cached SMC_003.wav
Cached SMC_004.wav
Cached SMC_005.wav
Cached SMC_006.wav
Processing SMC_009.wavProcessing SMC_007.wav
Processing SMC_008.wav
Processing SMC_010.wavProcessing SMC_012.wavProcessing SMC_011.wav
Processing SMC_013.wavProcessing SMC_015.wavProcessing SMC_014.wav
Processing SMC_016.wavProcessing SMC_018.wavProcessing SMC_017.wav
Processing SMC_019.wavProcessing SMC_022.wavProcessing SMC_021.wav
Processing SMC_023.wavProcessing SMC_026.wavProcessing SMC_024.wav
Processing SMC_027.wavProcessing SMC_030.wavProcessing SMC_028.wav
Processing SMC_032.wavProcessing SMC_034.wavProcessing SMC_033.wav
Processing SMC_036.wavProcessing SMC_037.wavProcessing SMC_035.wav
Processing SMC_041.wavProcessing SMC_043.wavProcessing SMC_038.wav
Processing SMC_044.wavProcessing SMC_047.wavProcessing SMC_042.wav
Processing SMC_048.wavProcessing SMC_052.wavProcessing SMC_046.wav
Processing SMC_054.wavProcessing SMC_056.wavProcessing SMC_051.wav
Processing SMC_057.wavProcessing SMC_059.wavProcessing SMC_055.wav
Processing SMC_060.wavProcessing SMC_061.wavProcessing SMC_058.wav
Processing SMC_065.wavProcessing SMC_064.wavProcessing SMC_063.wav
Processing SMC_069.wavProcessing SMC_067.wavProcessing SMC_066.wav
Processing SMC_073.wavProcessing SMC_071.wavProcessing SMC_068.wav
Processing SMC_076.wavProcessing SMC_074.wavProcessing SMC_072.wav
Processing SMC_084.wavProcessing SMC_079.wavProcessing SMC_075.wav
Processing SMC_087.wavProcessing SMC_082.wavProcessing SMC_080.wav
Processing SMC_092.wavProcessing SMC_085.wavProcessing SMC_086.wav
Processing SMC_096.wavProcessing SMC_089.wavProcessing SMC_088.wav
Processing SMC_100.wavProcessing SMC_095.wavProcessing SMC_093.wav
Processing SMC_105.wavProcessing SMC_098.wavProcessing SMC_099.wav
Processing SMC_111.wavProcessing SMC_101.wavProcessing SMC_103.wav
Processing SMC_113.wavProcessing SMC_104.wavProcessing SMC_106.wav
Processing SMC_117.wavProcessing SMC_109.wavProcessing SMC_114.wav
Processing SMC_120.wavProcessing SMC_116.wavProcessing SMC_118.wav
Processing SMC_126.wavProcessing SMC_119.wavProcessing SMC_121.wav
Processing SMC_135.wavProcessing SMC_124.wavProcessing SMC_127.wav
Processing SMC_137.wavProcessing SMC_130.wavProcessing SMC_133.wav
Processing SMC_142.wavProcessing SMC_139.wavProcessing SMC_140.wav
Processing SMC_147.wavProcessing SMC_143.wavProcessing SMC_146.wav
Processing SMC_150.wavProcessing SMC_149.wavProcessing SMC_148.wav
Processing SMC_151.wavProcessing SMC_152.wavProcessing SMC_153.wav
Processing SMC_154.wavProcessing SMC_158.wavProcessing SMC_157.wav
Processing SMC_161.wavProcessing SMC_166.wavProcessing SMC_159.wav
Processing SMC_168.wavProcessing SMC_169.wavProcessing SMC_167.wav
Processing SMC_171.wavProcessing SMC_172.wavProcessing SMC_170.wav
Processing SMC_174.wavProcessing SMC_175.wavProcessing SMC_173.wav
Processing SMC_178.wavProcessing SMC_179.wavProcessing SMC_176.wav
Processing SMC_182.wavProcessing SMC_184.wavProcessing SMC_181.wav
Processing SMC_188.wavProcessing SMC_190.wavProcessing SMC_187.wav
Processing SMC_194.wavProcessing SMC_193.wavProcessing SMC_192.wav
Processing SMC_198.wavProcessing SMC_195.wavProcessing SMC_197.wav
Processing SMC_203.wavProcessing SMC_199.wavProcessing SMC_202.wav
Processing SMC_205.wavProcessing SMC_204.wavProcessing SMC_206.wav
Processing SMC_207.wavProcessing SMC_208.wavProcessing SMC_209.wav
Processing SMC_212.wavProcessing SMC_211.wavProcessing SMC_213.wav
Processing SMC_215.wavProcessing SMC_214.wavProcessing SMC_216.wav
Processing SMC_217.wavProcessing SMC_219.wavProcessing SMC_222.wav
Processing SMC_221.wavProcessing SMC_220.wavProcessing SMC_224.wav
Processing SMC_225.wavProcessing SMC_223.wavProcessing SMC_229.wav
Processing SMC_227.wavProcessing SMC_226.wavProcessing SMC_236.wav
Processing SMC_235.wavProcessing SMC_232.wavProcessing SMC_239.wav
Processing SMC_241.wavProcessing SMC_237.wavProcessing SMC_243.wav
Processing SMC_244.wavProcessing SMC_242.wavProcessing SMC_248.wav
Processing SMC_252.wavProcessing SMC_249.wavProcessing SMC_251.wav
Processing SMC_256.wavProcessing SMC_253.wavProcessing SMC_254.wav
Processing SMC_257.wavProcessing SMC_255.wavProcessing SMC_259.wav
Processing SMC_261.wavProcessing SMC_258.wavProcessing SMC_262.wav
Processing SMC_264.wavProcessing SMC_260.wavProcessing SMC_265.wav
Processing SMC_269.wavProcessing SMC_263.wavProcessing SMC_271.wav
Processing SMC_273.wavProcessing SMC_266.wavProcessing SMC_274.wav
Processing SMC_276.wavProcessing SMC_272.wavProcessing SMC_277.wav
Processing SMC_279.wavProcessing SMC_275.wavProcessing SMC_280.wav
Processing SMC_282.wavProcessing SMC_278.wavProcessing SMC_283.wav
Processing SMC_285.wavProcessing SMC_281.wavProcessing SMC_286.wav
Processing SMC_288.wavProcessing SMC_284.wavProcessing SMC_289.wav
Processing SMC_287.wav
In [14]:
smc_results.to_json('/home/bmcfee/git/librosa_parameters/smc_log_results.json', orient='records')
In [12]:
smc_results = pd.read_json('/home/bmcfee/git/librosa_parameters/smc_log_results.json', orient='records')
In [13]:
smc_scores = smc_results.groupby(['aggregate', 'fmax', 'n_mels', 'ac_size', 'std_bpm', 'tightness']).mean()
In [14]:
best_igain = smc_scores['Information gain'].argmax()
In [15]:
best_fmeas = smc_scores['F-measure'].argmax()
In [16]:
best_amlt = smc_scores['Any Metric Level Total'].argmax()
In [17]:
best_cmlt = smc_scores['Correct Metric Level Total'].argmax()
In [18]:
smc_scores.loc[[best_igain, best_fmeas, best_amlt, best_cmlt]]
Out[18]:
Any Metric Level Continuous
Any Metric Level Total
Cemgil
Cemgil Best Metric Level
Correct Metric Level Continuous
Correct Metric Level Total
F-measure
Goto
Information gain
P-score
index
aggregate
fmax
n_mels
ac_size
std_bpm
tightness
median
8000
128
8
2
100
0.173
0.316
0.238
0.305
0.105
0.172
0.353
0.078
0.176
0.480
108
mean
8000
128
2
2
50
0.137
0.315
0.249
0.321
0.072
0.138
0.366
0.055
0.164
0.476
108
median
8000
128
8
1
50
0.155
0.334
0.243
0.303
0.097
0.172
0.361
0.078
0.174
0.493
108
2
1
100
0.163
0.328
0.240
0.299
0.107
0.177
0.356
0.083
0.172
0.493
108
In [19]:
smc_scores.loc[best_igain]
Out[19]:
Any Metric Level Continuous 0.173
Any Metric Level Total 0.316
Cemgil 0.238
Cemgil Best Metric Level 0.305
Correct Metric Level Continuous 0.105
Correct Metric Level Total 0.172
F-measure 0.353
Goto 0.078
Information gain 0.176
P-score 0.480
index 108.000
Name: (median, 8000, 128, 8, 2.0, 100), dtype: float64
In [20]:
smc_scores.loc[best_fmeas]
Out[20]:
Any Metric Level Continuous 0.137
Any Metric Level Total 0.315
Cemgil 0.249
Cemgil Best Metric Level 0.321
Correct Metric Level Continuous 0.072
Correct Metric Level Total 0.138
F-measure 0.366
Goto 0.055
Information gain 0.164
P-score 0.476
index 108.000
Name: (mean, 8000, 128, 2, 2.0, 50), dtype: float64
In [21]:
smc_scores.loc[best_amlt]
Out[21]:
Any Metric Level Continuous 0.155
Any Metric Level Total 0.334
Cemgil 0.243
Cemgil Best Metric Level 0.303
Correct Metric Level Continuous 0.097
Correct Metric Level Total 0.172
F-measure 0.361
Goto 0.078
Information gain 0.174
P-score 0.493
index 108.000
Name: (median, 8000, 128, 8, 1.0, 50), dtype: float64
In [22]:
smc_scores.loc[best_cmlt]
Out[22]:
Any Metric Level Continuous 0.163
Any Metric Level Total 0.328
Cemgil 0.240
Cemgil Best Metric Level 0.299
Correct Metric Level Continuous 0.107
Correct Metric Level Total 0.177
F-measure 0.356
Goto 0.083
Information gain 0.172
P-score 0.493
index 108.000
Name: (median, 8000, 128, 2, 1.0, 100), dtype: float64
In [12]:
beatles_data = make_beatles_corpus('/home/bmcfee/data/beatles_iso/')
In [13]:
beatles_results = p_analyze_corpus(beatles_data)
[Parallel(n_jobs=3)]: Done 1 out of 179 | elapsed: 6.4min remaining: 1138.7min
[Parallel(n_jobs=3)]: Done 19 out of 179 | elapsed: 24.7min remaining: 208.1min
[Parallel(n_jobs=3)]: Done 37 out of 179 | elapsed: 51.6min remaining: 197.9min
[Parallel(n_jobs=3)]: Done 55 out of 179 | elapsed: 73.6min remaining: 166.0min
[Parallel(n_jobs=3)]: Done 73 out of 179 | elapsed: 89.3min remaining: 129.7min
[Parallel(n_jobs=3)]: Done 91 out of 179 | elapsed: 104.8min remaining: 101.4min
[Parallel(n_jobs=3)]: Done 109 out of 179 | elapsed: 122.5min remaining: 78.7min
[Parallel(n_jobs=3)]: Done 127 out of 179 | elapsed: 148.8min remaining: 60.9min
[Parallel(n_jobs=3)]: Done 145 out of 179 | elapsed: 168.0min remaining: 39.4min
[Parallel(n_jobs=3)]: Done 163 out of 179 | elapsed: 193.1min remaining: 19.0min
[Parallel(n_jobs=3)]: Done 179 out of 179 | elapsed: 211.2min finished
Skipping line 286: expected 2 fields, saw 3
Processing 01_-_I_Saw_Her_Standing_There.flac
Processing 02_-_Misery.flac
Processing 03_-_Anna_(Go_To_Him).flac
Processing 06_-_Ask_Me_Why.flacProcessing 04_-_Chains.flacProcessing 05_-_Boys.flac
Processing 10_-_Baby_It's_You.flacProcessing 07_-_Please_Please_Me.flacProcessing 08_-_Love_Me_Do.flac
Processing 13_-_There's_A_Place.flacProcessing 09_-_P._S._I_Love_You.flacProcessing 12_-_A_Taste_Of_Honey.flac
Processing 02_-_All_I've_Got_To_Do.flacProcessing 11_-_Do_You_Want_To_Know_A_Secret.flacProcessing 01_-_It_Won't_Be_Long.flac
Processing 04_-_Don't_Bother_Me.flacProcessing 14_-_Twist_And_Shout.flacProcessing 03_-_All_My_Loving.flac
Processing 08_-_Roll_Over_Beethoven.flacProcessing 05_-_Little_Child.flacProcessing 07_-_Please_Mister_Postman.flac
Processing 11_-_I_Wanna_Be_Your_Man.flacProcessing 06_-_Till_There_Was_You.flacProcessing 10_-_You_Really_Got_A_Hold_On_Me.flac
Processing 13_-_Not_A_Second_Time.flacProcessing 09_-_Hold_Me_Tight.flacProcessing 14_-_Money.flac
Processing 02_-_I_Should_Have_Known_Better.flacProcessing 12_-_Devil_In_Her_Heart.flacProcessing 03_-_If_I_Fell.flac
Processing 05_-_And_I_Love_Her.flacProcessing 01_-_A_Hard_Day's_Night.flacProcessing 07_-_Can't_Buy_Me_Love.flac
Processing 08_-_Any_Time_At_All.flacProcessing 04_-_I'm_Happy_Just_To_Dance_With_You.flacProcessing 10_-_Things_We_Said_Today.flac
Processing 12_-_You_Can't_Do_That.flacProcessing 06_-_Tell_Me_Why.flacProcessing 13_-_I'll_Be_Back.flac
Processing 02_-_I'm_a_Loser.flacProcessing 09_-_I'll_Cry_Instead.flacProcessing 03_-_Baby's_In_Black.flac
Cached 05_-_I'll_Follow_the_Sun.flacProcessing 11_-_When_I_Get_Home.flacProcessing 07_-_Kansas_City-_Hey,_Hey,_Hey,_Hey.flac
Processing 06_-_Mr._Moonlight.flac
Processing 01_-_No_Reply.flacProcessing 10_-_Honey_Don't.flacProcessing 09_-_Words_of_Love.flac
Processing 04_-_Rock_and_Roll_Music.flacProcessing 13_-_What_You're_Doing.flacProcessing 12_-_I_Don't_Want_to_Spoil_the_Party.flac
Processing 08_-_Eight_Days_a_Week.flacProcessing 03_-_You've_Got_To_Hide_Your_Love_Away.flacProcessing 01_-_Help!.flac
Processing 11_-_Every_Little_Thing.flacProcessing 05_-_Another_Girl.flacProcessing 04_-_I_Need_You.flac
Processing 14_-_Everybody's_Trying_to_Be_My_Baby.flacProcessing 08_-_Act_Naturally.flacProcessing 07_-_Ticket_To_Ride.flac
Processing 02_-_The_Night_Before.flacProcessing 10_-_You_Like_Me_Too_Much.flacProcessing 12_-_I've_Just_Seen_a_Face.flac
Processing 06_-_You're_Going_To_Lose_That_Girl.flacProcessing 13_-_Yesterday.flacProcessing 01_-_Drive_My_Car.flac
Processing 09_-_It's_Only_Love.flacProcessing 02_-_Norwegian_Wood_(This_Bird_Has_Flown).flacProcessing 05_-_Think_For_Yourself.flac
Processing 11_-_Tell_Me_What_You_See.flacProcessing 04_-_Nowhere_Man.flacProcessing 06_-_The_Word.flac
Processing 14_-_Dizzy_Miss_Lizzy.flacProcessing 07_-_Michelle.flacProcessing 09_-_Girl.flac
Processing 03_-_You_Won't_See_Me.flacProcessing 10_-_I'm_Looking_Through_You.flacProcessing 13_-_If_I_Needed_Someone.flac
Processing 08_-_What_Goes_On.flacProcessing 12_-_Wait.flacProcessing 03_-_I'm_Only_Sleeping.flac
Processing 11_-_In_My_Life.flacProcessing 01_-_Taxman.flacProcessing 05_-_Here,_There_And_Everywhere.flac
Processing 14_-_Run_For_Your_Life.flacProcessing 06_-_Yellow_Submarine.flacProcessing 07_-_She_Said_She_Said.flac
Processing 02_-_Eleanor_Rigby.flacProcessing 09_-_And_Your_Bird_Can_Sing.flacProcessing 11_-_Doctor_Robert.flac
Processing 04_-_Love_You_To.flacProcessing 10_-_For_No_One.flacProcessing 14_-_Tomorrow_Never_Knows.flac
Processing 08_-_Good_Day_Sunshine.flacProcessing 13_-_Got_To_Get_You_Into_My_Life.flacProcessing 04_-_Getting_Better.flac
Processing 12_-_I_Want_To_Tell_You.flacProcessing 02_-_With_A_Little_Help_From_My_Friends.flacProcessing 06_-_She's_Leaving_Home.flac
Processing 01_-_Sgt._Pepper's_Lonely_Hearts_Club_Band.flacProcessing 05_-_Fixing_A_Hole.flacProcessing 10_-_Lovely_Rita.flac
Processing 03_-_Lucy_In_The_Sky_With_Diamonds.flacProcessing 08_-_Within_You_Without_You.flacProcessing 12_-_Sgt._Pepper's_Lonely_Hearts_Club_Band_(Reprise).flac
Processing 07_-_Being_For_The_Benefit_Of_Mr._Kite!.flacProcessing 02_-_The_Fool_On_The_Hill.flacProcessing 01_-_Magical_Mystery_Tour.flac
Processing 09_-_When_I'm_Sixty-Four.flacProcessing 04_-_Blue_Jay_Way.flacProcessing 03_-_Flying.flac
Processing 11_-_Good_Morning_Good_Morning.flacProcessing 07_-_Hello_Goodbye.flacProcessing 05_-_Your_Mother_Should_Know.flac
Processing 13_-_A_Day_In_The_Life.flacProcessing 09_-_Penny_Lane.flacProcessing 06_-_I_Am_The_Walrus.flac
Processing 08_-_Strawberry_Fields_Forever.flacProcessing CD1_-_02_-_Dear_Prudence.flacProcessing 10_-_Baby_You're_A_Rich_Man.flac
Processing 11_-_All_You_Need_Is_Love.flacProcessing CD1_-_07_-_While_My_Guitar_Gently_Weeps.flacProcessing CD1_-_01_-_Back_in_the_USSR.flac
Processing CD1_-_04_-_Ob-La-Di,_Ob-La-Da.flacProcessing CD1_-_12_-_Piggies.flacProcessing CD1_-_03_-_Glass_Onion.flac
Processing CD1_-_08_-_Happiness_is_a_Warm_Gun.flacProcessing CD1_-_15_-_Why_Don't_We_Do_It_In_The_Road.flacProcessing CD1_-_05_-_Wild_Honey_Pie.flac
Processing CD1_-_11_-_Black_Bird.flacProcessing CD1_-_16_-_I_Will.flacProcessing CD1_-_06_-_The_Continuing_Story_of_Bungalow_Bill.flac
Processing CD1_-_13_-_Rocky_Raccoon.flacProcessing CD1_-_17_-_Julia.flacProcessing CD1_-_09_-_Martha_My_Dear.flac
Processing CD2_-_01_-_Birthday.flacProcessing CD2_-_03_-_Mother_Nature's_Son.flacProcessing CD1_-_10_-_I'm_So_Tired.flac
Processing CD2_-_04_-_Everybody's_Got_Something_To_Hide_Except_Me_and_My_Monkey.flacProcessing CD2_-_06_-_Helter_Skelter.flacProcessing CD1_-_14_-_Don't_Pass_Me_By.flac
Processing CD2_-_05_-_Sexy_Sadie.flacProcessing CD2_-_10_-_Savoy_Truffle.flacProcessing CD2_-_02_-_Yer_Blues.flac
Processing CD2_-_08_-_Revolution_1.flacProcessing CD2_-_13_-_Good_Night.flacProcessing CD2_-_07_-_Long_Long_Long.flac
Processing 01_-_Come_Together.flacProcessing 04_-_Oh!_Darling.flacProcessing CD2_-_09_-_Honey_Pie.flac
Processing 05_-_Octopus's_Garden.flacProcessing 07_-_Here_Comes_The_Sun.flacProcessing CD2_-_11_-_Cry_Baby_Cry.flac
Processing 08_-_Because.flacProcessing 09_-_You_Never_Give_Me_Your_Money.flacProcessing 02_-_Something.flac
Processing 10_-_Sun_King.flacProcessing 14_-_Golden_Slumbers.flacProcessing 03_-_Maxwell's_Silver_Hammer.flac
Processing 11_-_Mean_Mr_Mustard.flacProcessing 15_-_Carry_That_Weight.flacProcessing 06_-_I_Want_You.flac
Processing 12_-_Polythene_Pam.flacProcessing 17_-_Her_Majesty.flacProcessing 05_-_Dig_It.flac
Processing 13_-_She_Came_In_Through_The_Bathroom_Window.flacProcessing 01_-_Two_of_Us.flacProcessing 06_-_Let_It_Be.flac
Processing 16_-_The_End.flacProcessing 03_-_Across_the_Universe.flacProcessing 10_-_The_Long_and_Winding_Road.flac
Processing 02_-_Dig_a_Pony.flacProcessing 09_-_One_After_909.flac
Processing 12_-_Get_Back.flacProcessing 04_-_I_Me_Mine.flac
Processing 07_-_Maggie_Mae.flac
Processing 08_-_I've_Got_A_Feeling.flac
Processing 11_-_For_You_Blue.flac
In [17]:
beatles_results.to_json('/home/bmcfee/git/librosa_parameters/beatles_beat_results.json', orient='records')
In [23]:
beatles_results = pd.read_json('/home/bmcfee/git/librosa_parameters/beatles_beat_results.json', orient='records')
In [24]:
beatles_scores = beatles_results.groupby(['aggregate', 'fmax', 'n_mels', 'ac_size', 'std_bpm', 'tightness']).mean()
In [25]:
beatles_scores
Out[25]:
Any Metric Level Continuous
Any Metric Level Total
Cemgil
Cemgil Best Metric Level
Correct Metric Level Continuous
Correct Metric Level Total
F-measure
Goto
Information gain
P-score
index
aggregate
fmax
n_mels
ac_size
std_bpm
tightness
mean
8000
32
2
0.5
50
0.492
0.722
0.536
0.602
0.432
0.578
0.690
0.503
0.428
0.714
89
100
0.575
0.735
0.543
0.613
0.494
0.588
0.698
0.520
0.450
0.720
89
400
0.591
0.731
0.542
0.611
0.513
0.591
0.694
0.525
0.445
0.721
89
1.0
50
0.530
0.783
0.556
0.627
0.463
0.628
0.719
0.536
0.454
0.743
89
100
0.620
0.802
0.567
0.641
0.527
0.643
0.732
0.570
0.478
0.755
89
400
0.645
0.800
0.570
0.642
0.552
0.645
0.731
0.581
0.477
0.755
89
2.0
50
0.514
0.765
0.554
0.622
0.440
0.601
0.716
0.536
0.451
0.736
89
100
0.595
0.779
0.562
0.633
0.496
0.610
0.725
0.547
0.474
0.743
89
400
0.627
0.779
0.564
0.636
0.521
0.611
0.724
0.553
0.474
0.742
89
4
0.5
50
0.492
0.722
0.536
0.602
0.432
0.578
0.690
0.503
0.428
0.714
89
100
0.575
0.735
0.543
0.613
0.494
0.588
0.698
0.520
0.450
0.720
89
400
0.591
0.731
0.542
0.611
0.513
0.591
0.694
0.525
0.445
0.721
89
1.0
50
0.530
0.783
0.556
0.627
0.463
0.628
0.719
0.536
0.454
0.743
89
100
0.620
0.802
0.567
0.641
0.527
0.643
0.732
0.570
0.478
0.755
89
400
0.645
0.800
0.570
0.642
0.552
0.645
0.731
0.581
0.477
0.755
89
2.0
50
0.514
0.765
0.554
0.622
0.440
0.601
0.716
0.536
0.451
0.736
89
100
0.595
0.779
0.562
0.633
0.496
0.610
0.725
0.547
0.474
0.743
89
400
0.627
0.779
0.564
0.636
0.521
0.611
0.724
0.553
0.474
0.742
89
8
0.5
50
0.492
0.722
0.536
0.602
0.432
0.578
0.690
0.503
0.428
0.714
89
100
0.575
0.735
0.543
0.613
0.494
0.588
0.698
0.520
0.450
0.720
89
400
0.591
0.731
0.542
0.611
0.513
0.591
0.694
0.525
0.445
0.721
89
1.0
50
0.530
0.783
0.556
0.627
0.463
0.628
0.719
0.536
0.454
0.743
89
100
0.620
0.802
0.567
0.641
0.527
0.643
0.732
0.570
0.478
0.755
89
400
0.645
0.800
0.570
0.642
0.552
0.645
0.731
0.581
0.477
0.755
89
2.0
50
0.514
0.765
0.554
0.622
0.440
0.601
0.716
0.536
0.451
0.736
89
100
0.595
0.779
0.562
0.633
0.496
0.610
0.725
0.547
0.474
0.743
89
400
0.627
0.779
0.564
0.636
0.521
0.611
0.724
0.553
0.474
0.742
89
64
2
0.5
50
0.539
0.725
0.553
0.613
0.468
0.588
0.704
0.503
0.441
0.724
89
100
0.587
0.731
0.557
0.620
0.502
0.591
0.706
0.514
0.456
0.727
89
400
0.600
0.727
0.551
0.617
0.519
0.591
0.700
0.525
0.453
0.724
89
1.0
50
0.578
0.783
0.570
0.640
0.495
0.628
0.727
0.542
0.467
0.746
89
100
0.634
0.794
0.575
0.649
0.533
0.634
0.731
0.559
0.483
0.750
89
400
0.656
0.796
0.571
0.651
0.550
0.631
0.725
0.564
0.484
0.746
89
2.0
50
0.577
0.790
0.577
0.646
0.483
0.620
0.733
0.547
0.472
0.748
89
100
0.624
0.799
0.582
0.653
0.516
0.625
0.738
0.553
0.489
0.753
89
400
0.648
0.798
0.579
0.654
0.536
0.622
0.733
0.559
0.489
0.748
89
4
0.5
50
0.539
0.725
0.553
0.613
0.468
0.588
0.704
0.503
0.441
0.724
89
100
0.587
0.731
0.557
0.620
0.502
0.591
0.706
0.514
0.456
0.727
89
400
0.600
0.727
0.551
0.617
0.519
0.591
0.700
0.525
0.453
0.724
89
1.0
50
0.578
0.783
0.570
0.640
0.495
0.628
0.727
0.542
0.467
0.746
89
100
0.634
0.794
0.575
0.649
0.533
0.634
0.731
0.559
0.483
0.750
89
400
0.656
0.796
0.571
0.651
0.550
0.631
0.725
0.564
0.484
0.746
89
2.0
50
0.577
0.790
0.577
0.646
0.483
0.620
0.733
0.547
0.472
0.748
89
100
0.624
0.799
0.582
0.653
0.516
0.625
0.738
0.553
0.489
0.753
89
400
0.648
0.798
0.579
0.654
0.536
0.622
0.733
0.559
0.489
0.748
89
8
0.5
50
0.539
0.725
0.553
0.613
0.468
0.588
0.704
0.503
0.441
0.724
89
100
0.587
0.731
0.557
0.620
0.502
0.591
0.706
0.514
0.456
0.727
89
400
0.600
0.727
0.551
0.617
0.519
0.591
0.700
0.525
0.453
0.724
89
1.0
50
0.578
0.783
0.570
0.640
0.495
0.628
0.727
0.542
0.467
0.746
89
100
0.634
0.794
0.575
0.649
0.533
0.634
0.731
0.559
0.483
0.750
89
400
0.656
0.796
0.571
0.651
0.550
0.631
0.725
0.564
0.484
0.746
89
2.0
50
0.577
0.790
0.577
0.646
0.483
0.620
0.733
0.547
0.472
0.748
89
100
0.624
0.799
0.582
0.653
0.516
0.625
0.738
0.553
0.489
0.753
89
400
0.648
0.798
0.579
0.654
0.536
0.622
0.733
0.559
0.489
0.748
89
128
2
0.5
50
0.541
0.724
0.565
0.623
0.470
0.589
0.711
0.508
0.446
0.730
89
100
0.593
0.726
0.566
0.627
0.511
0.591
0.710
0.520
0.457
0.730
89
400
0.601
0.719
0.562
0.622
0.525
0.591
0.705
0.536
0.451
0.728
89
1.0
50
0.598
0.801
0.583
0.660
0.499
0.629
0.734
0.553
0.479
0.747
89
100
0.656
0.808
0.586
0.666
0.542
0.633
0.737
0.559
0.491
0.749
89
400
0.670
0.808
0.586
0.666
0.559
0.634
0.735
0.575
0.491
0.750
89
2.0
50
0.587
0.799
0.591
0.663
0.488
0.626
0.743
0.559
0.479
0.751
89
100
0.639
0.805
0.594
0.669
0.531
0.629
0.746
0.564
0.492
0.754
89
400
0.662
0.807
0.596
0.672
0.551
0.631
0.746
0.581
0.496
0.754
89
4
0.5
50
0.541
0.724
0.565
0.623
0.470
0.589
0.711
0.508
0.446
0.730
89
100
0.593
0.726
0.566
0.627
0.511
0.591
0.710
0.520
0.457
0.730
89
400
0.601
0.719
0.562
0.622
0.525
0.591
0.705
0.536
0.451
0.728
89
1.0
50
0.598
0.801
0.583
0.660
0.499
0.629
0.734
0.553
0.479
0.747
89
100
0.656
0.808
0.586
0.666
0.542
0.633
0.737
0.559
0.491
0.749
89
400
0.670
0.808
0.586
0.666
0.559
0.634
0.735
0.575
0.491
0.750
89
2.0
50
0.587
0.799
0.591
0.663
0.488
0.626
0.743
0.559
0.479
0.751
89
100
0.639
0.805
0.594
0.669
0.531
0.629
0.746
0.564
0.492
0.754
89
400
0.662
0.807
0.596
0.672
0.551
0.631
0.746
0.581
0.496
0.754
89
8
0.5
50
0.541
0.724
0.565
0.623
0.470
0.589
0.711
0.508
0.446
0.730
89
100
0.593
0.726
0.566
0.627
0.511
0.591
0.710
0.520
0.457
0.730
89
400
0.601
0.719
0.562
0.622
0.525
0.591
0.705
0.536
0.451
0.728
89
1.0
50
0.598
0.801
0.583
0.660
0.499
0.629
0.734
0.553
0.479
0.747
89
100
0.656
0.808
0.586
0.666
0.542
0.633
0.737
0.559
0.491
0.749
89
400
0.670
0.808
0.586
0.666
0.559
0.634
0.735
0.575
0.491
0.750
89
2.0
50
0.587
0.799
0.591
0.663
0.488
0.626
0.743
0.559
0.479
0.751
89
100
0.639
0.805
0.594
0.669
0.531
0.629
0.746
0.564
0.492
0.754
89
400
0.662
0.807
0.596
0.672
0.551
0.631
0.746
0.581
0.496
0.754
89
11025
32
2
0.5
50
0.490
0.729
0.541
0.607
0.427
0.582
0.696
0.514
0.430
0.717
89
100
0.575
0.742
0.550
0.618
0.488
0.592
0.703
0.520
0.451
0.725
89
400
0.606
0.742
0.547
0.619
0.515
0.594
0.700
0.531
0.452
0.725
89
1.0
50
0.511
0.778
0.556
0.629
0.439
0.613
0.715
0.536
0.451
0.736
89
100
0.610
0.797
0.564
0.643
0.502
0.623
0.723
0.547
0.474
0.742
89
400
0.647
0.799
0.563
0.646
0.533
0.624
0.719
0.559
0.476
0.741
89
2.0
50
0.488
0.761
0.551
0.626
0.403
0.566
0.709
0.508
0.446
0.716
89
100
0.574
0.776
0.561
0.638
0.461
0.577
0.719
0.514
0.469
0.726
89
400
0.615
0.777
0.561
0.642
0.493
0.577
0.716
0.525
0.471
0.724
89
4
0.5
50
0.490
0.729
0.541
0.607
0.427
0.582
0.696
0.514
0.430
0.717
89
100
0.575
0.742
0.550
0.618
0.488
0.592
0.703
0.520
0.451
0.725
89
400
0.606
0.742
0.547
0.619
0.515
0.594
0.700
0.531
0.452
0.725
89
1.0
50
0.511
0.778
0.556
0.629
0.439
0.613
0.715
0.536
0.451
0.736
89
100
0.610
0.797
0.564
0.643
0.502
0.623
0.723
0.547
0.474
0.742
89
400
0.647
0.799
0.563
0.646
0.533
0.624
0.719
0.559
0.476
0.741
89
2.0
50
0.488
0.761
0.551
0.626
0.403
0.566
0.709
0.508
0.446
0.716
89
100
0.574
0.776
0.561
0.638
0.461
0.577
0.719
0.514
0.469
0.726
89
400
0.615
0.777
0.561
0.642
0.493
0.577
0.716
0.525
0.471
0.724
89
8
0.5
50
0.490
0.729
0.541
0.607
0.427
0.582
0.696
0.514
0.430
0.717
89
100
0.575
0.742
0.550
0.618
0.488
0.592
0.703
0.520
0.451
0.725
89
400
0.606
0.742
0.547
0.619
0.515
0.594
0.700
0.531
0.452
0.725
89
1.0
50
0.511
0.778
0.556
0.629
0.439
0.613
0.715
0.536
0.451
0.736
89
100
0.610
0.797
0.564
0.643
0.502
0.623
0.723
0.547
0.474
0.742
89
400
0.647
0.799
0.563
0.646
0.533
0.624
0.719
0.559
0.476
0.741
89
2.0
50
0.488
0.761
0.551
0.626
0.403
0.566
0.709
0.508
0.446
0.716
89
100
0.574
0.776
0.561
0.638
0.461
0.577
0.719
0.514
0.469
0.726
89
400
0.615
0.777
0.561
0.642
0.493
0.577
0.716
0.525
0.471
0.724
89
64
2
0.5
50
0.525
0.734
0.559
0.619
0.456
0.593
0.710
0.514
0.442
0.729
89
100
0.587
0.741
0.562
0.626
0.499
0.597
0.713
0.520
0.457
0.732
89
400
0.609
0.735
0.557
0.622
0.525
0.596
0.706
0.536
0.455
0.729
89
1.0
50
0.566
0.793
0.572
0.645
0.481
0.634
0.730
0.553
0.468
0.748
89
100
0.628
0.803
0.578
0.653
0.525
0.641
0.735
0.559
0.484
0.754
89
400
0.654
0.801
0.575
0.652
0.550
0.638
0.729
0.575
0.484
0.750
89
2.0
50
0.546
0.782
0.574
0.645
0.452
0.599
0.732
0.536
0.465
0.740
89
100
0.603
0.790
0.579
0.652
0.493
0.605
0.737
0.536
0.482
0.745
89
400
0.636
0.792
0.579
0.654
0.524
0.604
0.734
0.553
0.485
0.742
89
4
0.5
50
0.525
0.734
0.559
0.619
0.456
0.593
0.710
0.514
0.442
0.729
89
100
0.587
0.741
0.562
0.626
0.499
0.597
0.713
0.520
0.457
0.732
89
400
0.609
0.735
0.557
0.622
0.525
0.596
0.706
0.536
0.455
0.729
89
1.0
50
0.566
0.793
0.572
0.645
0.481
0.634
0.730
0.553
0.468
0.748
89
100
0.628
0.803
0.578
0.653
0.525
0.641
0.735
0.559
0.484
0.754
89
400
0.654
0.801
0.575
0.652
0.550
0.638
0.729
0.575
0.484
0.750
89
2.0
50
0.546
0.782
0.574
0.645
0.452
0.599
0.732
0.536
0.465
0.740
89
100
0.603
0.790
0.579
0.652
0.493
0.605
0.737
0.536
0.482
0.745
89
400
0.636
0.792
0.579
0.654
0.524
0.604
0.734
0.553
0.485
0.742
89
8
0.5
50
0.525
0.734
0.559
0.619
0.456
0.593
0.710
0.514
0.442
0.729
89
100
0.587
0.741
0.562
0.626
0.499
0.597
0.713
0.520
0.457
0.732
89
400
0.609
0.735
0.557
0.622
0.525
0.596
0.706
0.536
0.455
0.729
89
1.0
50
0.566
0.793
0.572
0.645
0.481
0.634
0.730
0.553
0.468
0.748
89
100
0.628
0.803
0.578
0.653
0.525
0.641
0.735
0.559
0.484
0.754
89
400
0.654
0.801
0.575
0.652
0.550
0.638
0.729
0.575
0.484
0.750
89
2.0
50
0.546
0.782
0.574
0.645
0.452
0.599
0.732
0.536
0.465
0.740
89
100
0.603
0.790
0.579
0.652
0.493
0.605
0.737
0.536
0.482
0.745
89
400
0.636
0.792
0.579
0.654
0.524
0.604
0.734
0.553
0.485
0.742
89
128
2
0.5
50
0.539
0.728
0.567
0.624
0.469
0.594
0.713
0.514
0.446
0.732
89
100
0.593
0.730
0.568
0.628
0.510
0.595
0.712
0.525
0.457
0.732
89
400
0.615
0.725
0.565
0.625
0.538
0.597
0.709
0.542
0.456
0.733
89
1.0
50
0.590
0.800
0.581
0.658
0.492
0.628
0.733
0.547
0.476
0.746
89
100
0.648
0.806
0.584
0.664
0.535
0.631
0.735
0.559
0.488
0.748
89
400
0.679
0.807
0.585
0.665
0.563
0.634
0.734
0.575
0.491
0.750
89
2.0
50
0.578
0.804
0.588
0.662
0.478
0.622
0.741
0.553
0.477
0.750
89
100
0.630
0.808
0.591
0.668
0.518
0.624
0.742
0.553
0.489
0.751
89
400
0.665
0.810
0.593
0.671
0.550
0.626
0.743
0.570
0.493
0.752
89
4
0.5
50
0.539
0.728
0.567
0.624
0.469
0.594
0.713
0.514
0.446
0.732
89
100
0.593
0.730
0.568
0.628
0.510
0.595
0.712
0.525
0.457
0.732
89
400
0.615
0.725
0.565
0.625
0.538
0.597
0.709
0.542
0.456
0.733
89
1.0
50
0.590
0.800
0.581
0.658
0.492
0.628
0.733
0.547
0.476
0.746
89
100
0.648
0.806
0.584
0.664
0.535
0.631
0.735
0.559
0.488
0.748
89
400
0.679
0.807
0.585
0.665
0.563
0.634
0.734
0.575
0.491
0.750
89
2.0
50
0.578
0.804
0.588
0.662
0.478
0.622
0.741
0.553
0.477
0.750
89
100
0.630
0.808
0.591
0.668
0.518
0.624
0.742
0.553
0.489
0.751
89
400
0.665
0.810
0.593
0.671
0.550
0.626
0.743
0.570
0.493
0.752
89
8
0.5
50
0.539
0.728
0.567
0.624
0.469
0.594
0.713
0.514
0.446
0.732
89
100
0.593
0.730
0.568
0.628
0.510
0.595
0.712
0.525
0.457
0.732
89
400
0.615
0.725
0.565
0.625
0.538
0.597
0.709
0.542
0.456
0.733
89
1.0
50
0.590
0.800
0.581
0.658
0.492
0.628
0.733
0.547
0.476
0.746
89
100
0.648
0.806
0.584
0.664
0.535
0.631
0.735
0.559
0.488
0.748
89
400
0.679
0.807
0.585
0.665
0.563
0.634
0.734
0.575
0.491
0.750
89
2.0
50
0.578
0.804
0.588
0.662
0.478
0.622
0.741
0.553
0.477
0.750
89
100
0.630
0.808
0.591
0.668
0.518
0.624
0.742
0.553
0.489
0.751
89
400
0.665
0.810
0.593
0.671
0.550
0.626
0.743
0.570
0.493
0.752
89
median
8000
32
2
0.5
50
0.487
0.750
0.534
0.598
0.436
0.615
0.700
0.508
0.435
0.734
89
100
0.590
0.768
0.542
0.612
0.507
0.625
0.709
0.542
0.460
0.740
89
400
0.617
0.771
0.541
0.617
0.526
0.624
0.704
0.553
0.463
0.736
89
1.0
50
0.493
0.769
0.543
0.612
0.424
0.604
0.712
0.508
0.443
0.738
89
100
0.583
0.783
0.552
0.623
0.486
0.615
0.722
0.536
0.466
0.746
89
400
0.622
0.786
0.555
0.628
0.516
0.617
0.721
0.553
0.472
0.744
89
2.0
50
0.490
0.761
0.527
0.600
0.402
0.563
0.692
0.475
0.442
0.715
89
100
0.582
0.775
0.535
0.611
0.454
0.571
0.701
0.497
0.465
0.721
89
400
0.629
0.782
0.538
0.618
0.482
0.573
0.701
0.508
0.475
0.720
89
4
0.5
50
0.487
0.750
0.534
0.598
0.436
0.615
0.700
0.508
0.435
0.734
89
100
0.590
0.768
0.542
0.612
0.507
0.625
0.709
0.542
0.460
0.740
89
400
0.617
0.771
0.541
0.617
0.526
0.624
0.704
0.553
0.463
0.736
89
1.0
50
0.493
0.769
0.543
0.612
0.424
0.604
0.712
0.508
0.443
0.738
89
100
0.583
0.783
0.552
0.623
0.486
0.615
0.722
0.536
0.466
0.746
89
400
0.622
0.786
0.555
0.628
0.516
0.617
0.721
0.553
0.472
0.744
89
2.0
50
0.490
0.761
0.527
0.600
0.402
0.563
0.692
0.475
0.442
0.715
89
100
0.582
0.775
0.535
0.611
0.454
0.571
0.701
0.497
0.465
0.721
89
400
0.629
0.782
0.538
0.618
0.482
0.573
0.701
0.508
0.475
0.720
89
8
0.5
50
0.487
0.750
0.534
0.598
0.436
0.615
0.700
0.508
0.435
0.734
89
100
0.590
0.768
0.542
0.612
0.507
0.625
0.709
0.542
0.460
0.740
89
400
0.617
0.771
0.541
0.617
0.526
0.624
0.704
0.553
0.463
0.736
89
1.0
50
0.493
0.769
0.543
0.612
0.424
0.604
0.712
0.508
0.443
0.738
89
100
0.583
0.783
0.552
0.623
0.486
0.615
0.722
0.536
0.466
0.746
89
400
0.622
0.786
0.555
0.628
0.516
0.617
0.721
0.553
0.472
0.744
89
2.0
50
0.490
0.761
0.527
0.600
0.402
0.563
0.692
0.475
0.442
0.715
89
100
0.582
0.775
0.535
0.611
0.454
0.571
0.701
0.497
0.465
0.721
89
400
0.629
0.782
0.538
0.618
0.482
0.573
0.701
0.508
0.475
0.720
89
64
2
0.5
50
0.546
0.780
0.555
0.623
0.476
0.633
0.717
0.536
0.461
0.747
89
100
0.619
0.793
0.561
0.633
0.528
0.642
0.724
0.553
0.481
0.753
89
400
0.642
0.794
0.558
0.635
0.542
0.639
0.718
0.553
0.481
0.748
89
1.0
50
0.560
0.794
0.563
0.633
0.476
0.635
0.730
0.553
0.470
0.755
89
100
0.624
0.805
0.569
0.642
0.519
0.642
0.737
0.564
0.488
0.760
89
400
0.653
0.806
0.569
0.646
0.538
0.640
0.733
0.564
0.491
0.755
89
2.0
50
0.535
0.774
0.546
0.623
0.430
0.572
0.709
0.492
0.466
0.724
89
100
0.609
0.784
0.551
0.631
0.466
0.576
0.713
0.503
0.484
0.726
89
400
0.650
0.788
0.553
0.637
0.490
0.574
0.711
0.503
0.489
0.724
89
4
0.5
50
0.546
0.780
0.555
0.623
0.476
0.633
0.717
0.536
0.461
0.747
89
100
0.619
0.793
0.561
0.633
0.528
0.642
0.724
0.553
0.481
0.753
89
400
0.642
0.794
0.558
0.635
0.542
0.639
0.718
0.553
0.481
0.748
89
1.0
50
0.560
0.794
0.563
0.633
0.476
0.635
0.730
0.553
0.470
0.755
89
100
0.624
0.805
0.569
0.642
0.519
0.642
0.737
0.564
0.488
0.760
89
400
0.653
0.806
0.569
0.646
0.538
0.640
0.733
0.564
0.491
0.755
89
2.0
50
0.535
0.774
0.546
0.623
0.430
0.572
0.709
0.492
0.466
0.724
89
100
0.609
0.784
0.551
0.631
0.466
0.576
0.713
0.503
0.484
0.726
89
400
0.650
0.788
0.553
0.637
0.490
0.574
0.711
0.503
0.489
0.724
89
8
0.5
50
0.546
0.780
0.555
0.623
0.476
0.633
0.717
0.536
0.461
0.747
89
100
0.619
0.793
0.561
0.633
0.528
0.642
0.724
0.553
0.481
0.753
89
400
0.642
0.794
0.558
0.635
0.542
0.639
0.718
0.553
0.481
0.748
89
1.0
50
0.560
0.794
0.563
0.633
0.476
0.635
0.730
0.553
0.470
0.755
89
100
0.624
0.805
0.569
0.642
0.519
0.642
0.737
0.564
0.488
0.760
89
400
0.653
0.806
0.569
0.646
0.538
0.640
0.733
0.564
0.491
0.755
89
2.0
50
0.535
0.774
0.546
0.623
0.430
0.572
0.709
0.492
0.466
0.724
89
100
0.609
0.784
0.551
0.631
0.466
0.576
0.713
0.503
0.484
0.726
89
400
0.650
0.788
0.553
0.637
0.490
0.574
0.711
0.503
0.489
0.724
89
128
2
0.5
50
0.586
0.804
0.568
0.646
0.496
0.641
0.727
0.547
0.480
0.752
89
100
0.645
0.812
0.569
0.650
0.534
0.645
0.727
0.553
0.493
0.754
89
400
0.668
0.811
0.569
0.652
0.554
0.646
0.725
0.564
0.495
0.753
89
1.0
50
0.587
0.806
0.580
0.653
0.487
0.639
0.741
0.553
0.483
0.761
89
100
0.635
0.812
0.582
0.656
0.524
0.642
0.742
0.559
0.493
0.762
89
400
0.659
0.813
0.583
0.659
0.545
0.643
0.740
0.570
0.497
0.761
89
2.0
50
0.558
0.780
0.561
0.639
0.439
0.576
0.718
0.497
0.478
0.728
89
100
0.622
0.787
0.564
0.645
0.474
0.578
0.721
0.503
0.492
0.730
89
400
0.654
0.789
0.566
0.649
0.499
0.581
0.722
0.520
0.498
0.731
89
4
0.5
50
0.586
0.804
0.568
0.646
0.496
0.641
0.727
0.547
0.480
0.752
89
100
0.645
0.812
0.569
0.650
0.534
0.645
0.727
0.553
0.493
0.754
89
400
0.668
0.811
0.569
0.652
0.554
0.646
0.725
0.564
0.495
0.753
89
1.0
50
0.587
0.806
0.580
0.653
0.487
0.639
0.741
0.553
0.483
0.761
89
100
0.635
0.812
0.582
0.656
0.524
0.642
0.742
0.559
0.493
0.762
89
400
0.659
0.813
0.583
0.659
0.545
0.643
0.740
0.570
0.497
0.761
89
2.0
50
0.558
0.780
0.561
0.639
0.439
0.576
0.718
0.497
0.478
0.728
89
100
0.622
0.787
0.564
0.645
0.474
0.578
0.721
0.503
0.492
0.730
89
400
0.654
0.789
0.566
0.649
0.499
0.581
0.722
0.520
0.498
0.731
89
8
0.5
50
0.586
0.804
0.568
0.646
0.496
0.641
0.727
0.547
0.480
0.752
89
100
0.645
0.812
0.569
0.650
0.534
0.645
0.727
0.553
0.493
0.754
89
400
0.668
0.811
0.569
0.652
0.554
0.646
0.725
0.564
0.495
0.753
89
1.0
50
0.587
0.806
0.580
0.653
0.487
0.639
0.741
0.553
0.483
0.761
89
100
0.635
0.812
0.582
0.656
0.524
0.642
0.742
0.559
0.493
0.762
89
400
0.659
0.813
0.583
0.659
0.545
0.643
0.740
0.570
0.497
0.761
89
2.0
50
0.558
0.780
0.561
0.639
0.439
0.576
0.718
0.497
0.478
0.728
89
100
0.622
0.787
0.564
0.645
0.474
0.578
0.721
0.503
0.492
0.730
89
400
0.654
0.789
0.566
0.649
0.499
0.581
0.722
0.520
0.498
0.731
89
11025
32
2
0.5
50
0.487
0.756
0.536
0.606
0.426
0.607
0.700
0.520
0.438
0.731
89
100
0.594
0.774
0.545
0.620
0.503
0.617
0.709
0.542
0.462
0.738
89
400
0.617
0.774
0.544
0.621
0.521
0.618
0.706
0.547
0.463
0.736
89
1.0
50
0.492
0.768
0.546
0.615
0.417
0.603
0.711
0.531
0.443
0.737
89
100
0.591
0.784
0.555
0.629
0.487
0.612
0.721
0.542
0.467
0.744
89
400
0.623
0.782
0.555
0.630
0.513
0.614
0.718
0.547
0.470
0.742
89
2.0
50
0.477
0.756
0.528
0.603
0.384
0.551
0.692
0.486
0.443
0.710
89
100
0.571
0.771
0.537
0.616
0.441
0.558
0.702
0.492
0.466
0.716
89
400
0.624
0.774
0.539
0.620
0.476
0.561
0.701
0.497
0.473
0.716
89
4
0.5
50
0.487
0.756
0.536
0.606
0.426
0.607
0.700
0.520
0.438
0.731
89
100
0.594
0.774
0.545
0.620
0.503
0.617
0.709
0.542
0.462
0.738
89
400
0.617
0.774
0.544
0.621
0.521
0.618
0.706
0.547
0.463
0.736
89
1.0
50
0.492
0.768
0.546
0.615
0.417
0.603
0.711
0.531
0.443
0.737
89
100
0.591
0.784
0.555
0.629
0.487
0.612
0.721
0.542
0.467
0.744
89
400
0.623
0.782
0.555
0.630
0.513
0.614
0.718
0.547
0.470
0.742
89
2.0
50
0.477
0.756
0.528
0.603
0.384
0.551
0.692
0.486
0.443
0.710
89
100
0.571
0.771
0.537
0.616
0.441
0.558
0.702
0.492
0.466
0.716
89
400
0.624
0.774
0.539
0.620
0.476
0.561
0.701
0.497
0.473
0.716
89
8
0.5
50
0.487
0.756
0.536
0.606
0.426
0.607
0.700
0.520
0.438
0.731
89
100
0.594
0.774
0.545
0.620
0.503
0.617
0.709
0.542
0.462
0.738
89
400
0.617
0.774
0.544
0.621
0.521
0.618
0.706
0.547
0.463
0.736
89
1.0
50
0.492
0.768
0.546
0.615
0.417
0.603
0.711
0.531
0.443
0.737
89
100
0.591
0.784
0.555
0.629
0.487
0.612
0.721
0.542
0.467
0.744
89
400
0.623
0.782
0.555
0.630
0.513
0.614
0.718
0.547
0.470
0.742
89
2.0
50
0.477
0.756
0.528
0.603
0.384
0.551
0.692
0.486
0.443
0.710
89
100
0.571
0.771
0.537
0.616
0.441
0.558
0.702
0.492
0.466
0.716
89
400
0.624
0.774
0.539
0.620
0.476
0.561
0.701
0.497
0.473
0.716
89
64
2
0.5
50
0.544
0.791
0.551
0.627
0.468
0.631
0.714
0.542
0.464
0.743
89
100
0.627
0.802
0.556
0.636
0.523
0.638
0.720
0.553
0.481
0.748
89
400
0.649
0.803
0.555
0.637
0.539
0.638
0.716
0.559
0.484
0.746
89
1.0
50
0.547
0.794
0.563
0.635
0.458
0.628
0.728
0.553
0.467
0.752
89
100
0.616
0.802
0.567
0.641
0.508
0.634
0.733
0.559
0.481
0.755
89
400
0.646
0.806
0.568
0.645
0.532
0.635
0.731
0.564
0.488
0.754
89
2.0
50
0.529
0.771
0.545
0.621
0.426
0.572
0.706
0.497
0.463
0.722
89
100
0.609
0.780
0.550
0.628
0.469
0.576
0.710
0.503
0.480
0.725
89
400
0.646
0.785
0.553
0.634
0.496
0.578
0.712
0.514
0.489
0.726
89
4
0.5
50
0.544
0.791
0.551
0.627
0.468
0.631
0.714
0.542
0.464
0.743
89
100
0.627
0.802
0.556
0.636
0.523
0.638
0.720
0.553
0.481
0.748
89
400
0.649
0.803
0.555
0.637
0.539
0.638
0.716
0.559
0.484
0.746
89
1.0
50
0.547
0.794
0.563
0.635
0.458
0.628
0.728
0.553
0.467
0.752
89
100
0.616
0.802
0.567
0.641
0.508
0.634
0.733
0.559
0.481
0.755
89
400
0.646
0.806
0.568
0.645
0.532
0.635
0.731
0.564
0.488
0.754
89
2.0
50
0.529
0.771
0.545
0.621
0.426
0.572
0.706
0.497
0.463
0.722
89
100
0.609
0.780
0.550
0.628
0.469
0.576
0.710
0.503
0.480
0.725
89
400
0.646
0.785
0.553
0.634
0.496
0.578
0.712
0.514
0.489
0.726
89
8
0.5
50
0.544
0.791
0.551
0.627
0.468
0.631
0.714
0.542
0.464
0.743
89
100
0.627
0.802
0.556
0.636
0.523
0.638
0.720
0.553
0.481
0.748
89
400
0.649
0.803
0.555
0.637
0.539
0.638
0.716
0.559
0.484
0.746
89
1.0
50
0.547
0.794
0.563
0.635
0.458
0.628
0.728
0.553
0.467
0.752
89
100
0.616
0.802
0.567
0.641
0.508
0.634
0.733
0.559
0.481
0.755
89
400
0.646
0.806
0.568
0.645
0.532
0.635
0.731
0.564
0.488
0.754
89
2.0
50
0.529
0.771
0.545
0.621
0.426
0.572
0.706
0.497
0.463
0.722
89
100
0.609
0.780
0.550
0.628
0.469
0.576
0.710
0.503
0.480
0.725
89
400
0.646
0.785
0.553
0.634
0.496
0.578
0.712
0.514
0.489
0.726
89
128
2
0.5
50
0.582
0.801
0.565
0.642
0.493
0.643
0.724
0.553
0.477
0.750
89
100
0.645
0.812
0.570
0.650
0.534
0.650
0.729
0.559
0.492
0.756
89
400
0.677
0.813
0.570
0.652
0.564
0.651
0.727
0.570
0.495
0.755
89
1.0
50
0.570
0.801
0.570
0.648
0.471
0.630
0.731
0.553
0.477
0.751
89
100
0.630
0.810
0.575
0.654
0.511
0.636
0.735
0.553
0.491
0.756
89
400
0.663
0.811
0.576
0.657
0.544
0.637
0.734
0.564
0.496
0.754
89
2.0
50
0.551
0.779
0.557
0.635
0.436
0.581
0.715
0.503
0.474
0.729
89
100
0.620
0.786
0.562
0.641
0.475
0.584
0.719
0.503
0.489
0.733
89
400
0.662
0.789
0.565
0.646
0.512
0.587
0.720
0.520
0.494
0.733
89
4
0.5
50
0.582
0.801
0.565
0.642
0.493
0.643
0.724
0.553
0.477
0.750
89
100
0.645
0.812
0.570
0.650
0.534
0.650
0.729
0.559
0.492
0.756
89
400
0.677
0.813
0.570
0.652
0.564
0.651
0.727
0.570
0.495
0.755
89
1.0
50
0.570
0.801
0.570
0.648
0.471
0.630
0.731
0.553
0.477
0.751
89
100
0.630
0.810
0.575
0.654
0.511
0.636
0.735
0.553
0.491
0.756
89
400
0.663
0.811
0.576
0.657
0.544
0.637
0.734
0.564
0.496
0.754
89
2.0
50
0.551
0.779
0.557
0.635
0.436
0.581
0.715
0.503
0.474
0.729
89
100
0.620
0.786
0.562
0.641
0.475
0.584
0.719
0.503
0.489
0.733
89
400
0.662
0.789
0.565
0.646
0.512
0.587
0.720
0.520
0.494
0.733
89
8
0.5
50
0.582
0.801
0.565
0.642
0.493
0.643
0.724
0.553
0.477
0.750
89
100
0.645
0.812
0.570
0.650
0.534
0.650
0.729
0.559
0.492
0.756
89
400
0.677
0.813
0.570
0.652
0.564
0.651
0.727
0.570
0.495
0.755
89
1.0
50
0.570
0.801
0.570
0.648
0.471
0.630
0.731
0.553
0.477
0.751
89
100
0.630
0.810
0.575
0.654
0.511
0.636
0.735
0.553
0.491
0.756
89
400
0.663
0.811
0.576
0.657
0.544
0.637
0.734
0.564
0.496
0.754
89
2.0
50
0.551
0.779
0.557
0.635
0.436
0.581
0.715
0.503
0.474
0.729
89
100
0.620
0.786
0.562
0.641
0.475
0.584
0.719
0.503
0.489
0.733
89
400
0.662
0.789
0.565
0.646
0.512
0.587
0.720
0.520
0.494
0.733
89
In [26]:
best_igain_b = beatles_scores['Information gain'].argmax()
best_fmeas_b = beatles_scores['F-measure'].argmax()
best_amlt_b = beatles_scores['Any Metric Level Total'].argmax()
best_cmlt_b = beatles_scores['Correct Metric Level Total'].argmax()
In [27]:
print best_igain
print best_fmeas
print best_amlt
print best_cmlt
print
print best_igain_b
print best_fmeas_b
print best_amlt_b
print best_cmlt_b
(u'median', 8000, 128, 8, 2.0, 100)
(u'mean', 8000, 128, 2, 2.0, 50)
(u'median', 8000, 128, 8, 1.0, 50)
(u'median', 8000, 128, 2, 1.0, 100)
(u'median', 8000, 128, 2, 2.0, 400)
(u'mean', 8000, 128, 2, 2.0, 100)
(u'median', 8000, 128, 2, 1.0, 400)
(u'median', 11025, 128, 2, 0.5, 400)
In [28]:
good_scores = ['Any Metric Level Total', 'Information gain', 'F-measure', 'Correct Metric Level Total']
In [29]:
smc_scores.loc[[best_igain, best_fmeas, best_amlt]][good_scores]
Out[29]:
Any Metric Level Total
Information gain
F-measure
Correct Metric Level Total
aggregate
fmax
n_mels
ac_size
std_bpm
tightness
median
8000
128
8
2
100
0.316
0.176
0.353
0.172
mean
8000
128
2
2
50
0.315
0.164
0.366
0.138
median
8000
128
8
1
50
0.334
0.174
0.361
0.172
In [30]:
smc_scores.loc[[best_igain_b, best_fmeas_b, best_amlt_b, best_cmlt_b]][good_scores]
Out[30]:
Any Metric Level Total
Information gain
F-measure
Correct Metric Level Total
aggregate
fmax
n_mels
ac_size
std_bpm
tightness
median
8000
128
2
2.0
400
0.298
0.161
0.326
0.155
mean
8000
128
2
2.0
100
0.312
0.164
0.357
0.145
median
8000
128
2
1.0
400
0.301
0.160
0.329
0.161
11025
128
2
0.5
400
0.264
0.147
0.306
0.124
In [31]:
beatles_scores.loc[[best_igain, best_fmeas, best_amlt, best_cmlt]][good_scores]
Out[31]:
Any Metric Level Total
Information gain
F-measure
Correct Metric Level Total
aggregate
fmax
n_mels
ac_size
std_bpm
tightness
median
8000
128
8
2
100
0.787
0.492
0.721
0.578
mean
8000
128
2
2
50
0.799
0.479
0.743
0.626
median
8000
128
8
1
50
0.806
0.483
0.741
0.639
2
1
100
0.812
0.493
0.742
0.642
In [32]:
beatles_scores.loc[[best_igain_b, best_fmeas_b, best_amlt_b, best_cmlt_b]][good_scores]
Out[32]:
Any Metric Level Total
Information gain
F-measure
Correct Metric Level Total
aggregate
fmax
n_mels
ac_size
std_bpm
tightness
median
8000
128
2
2.0
400
0.789
0.498
0.722
0.581
mean
8000
128
2
2.0
100
0.805
0.492
0.746
0.629
median
8000
128
2
1.0
400
0.813
0.497
0.740
0.643
11025
128
2
0.5
400
0.813
0.495
0.727
0.651
In [37]:
beatles_results['fmax'].unique()
Out[37]:
array([ 8000, 11025])
In [33]:
smc_scores.loc[[('mean', 11025, 128, 4, 1.0, 400),
best_amlt,
best_amlt_b,
best_cmlt,
best_cmlt_b,
('median', 8000, 128, 4, 1.0, 100),]][good_scores]
Out[33]:
Any Metric Level Total
Information gain
F-measure
Correct Metric Level Total
aggregate
fmax
n_mels
ac_size
std_bpm
tightness
mean
11025
128
4
1.0
400
0.282
0.146
0.313
0.109
median
8000
128
8
1.0
50
0.334
0.174
0.361
0.172
2
1.0
400
0.301
0.160
0.329
0.161
100
0.328
0.172
0.356
0.177
11025
128
2
0.5
400
0.264
0.147
0.306
0.124
8000
128
4
1.0
100
0.329
0.172
0.356
0.177
In [35]:
beatles_scores.loc[[(u'mean', 11025.0, 128, 4, 1.0, 400),
best_amlt,
best_amlt_b,
best_cmlt,
best_cmlt_b,
('median', 8000, 128, 4, 1.0, 100),]][good_scores]
Out[35]:
Any Metric Level Total
Information gain
F-measure
Correct Metric Level Total
aggregate
fmax
n_mels
ac_size
std_bpm
tightness
mean
11025
128
4
1.0
400
0.807
0.491
0.734
0.634
median
8000
128
8
1.0
50
0.806
0.483
0.741
0.639
2
1.0
400
0.813
0.497
0.740
0.643
100
0.812
0.493
0.742
0.642
11025
128
2
0.5
400
0.813
0.495
0.727
0.651
8000
128
4
1.0
100
0.812
0.493
0.742
0.642
In [36]:
best_igain_b
Out[36]:
(u'median', 8000, 128, 2, 2.0, 400)
In [37]:
all_results = pd.concat([smc_results, beatles_results])
In [38]:
all_scores = all_results.groupby(['aggregate', 'fmax', 'n_mels', 'ac_size', 'std_bpm', 'tightness']).mean()
In [39]:
best_igain_a = all_scores['Information gain'].argmax()
best_fmeas_a = all_scores['F-measure'].argmax()
best_amlt_a = all_scores['Any Metric Level Total'].argmax()
best_cmlt_a = all_scores['Correct Metric Level Total'].argmax()
In [40]:
all_scores.loc[[(u'mean', 11025.0, 128, 4, 1.0, 400),
best_cmlt,
best_cmlt_b,
best_cmlt_a,
('median', 8000, 128, 4, 1.0, 100),]]
Out[40]:
Any Metric Level Continuous
Any Metric Level Total
Cemgil
Cemgil Best Metric Level
Correct Metric Level Continuous
Correct Metric Level Total
F-measure
Goto
Information gain
P-score
index
aggregate
fmax
n_mels
ac_size
std_bpm
tightness
mean
11025
128
4
1.0
400
0.385
0.519
0.381
0.455
0.292
0.346
0.504
0.295
0.302
0.586
99.412
median
8000
128
2
1.0
100
0.376
0.547
0.395
0.461
0.295
0.387
0.530
0.298
0.317
0.615
99.412
11025
128
2
0.5
400
0.377
0.512
0.371
0.443
0.295
0.362
0.496
0.295
0.304
0.590
99.412
8000
128
2
1.0
100
0.376
0.547
0.395
0.461
0.295
0.387
0.530
0.298
0.317
0.615
99.412
4
1.0
100
0.377
0.547
0.395
0.461
0.295
0.387
0.530
0.298
0.317
0.615
99.412
Content source: bmcfee/librosa_parameters
Similar notebooks: